A Real-Time And Reliable Forwarding Scheme Of Publish-Subscribe System Based On WSN

Topic-based publish/subscribe system is widely used in IoT (Internet of Things). But many existing systems cannot work efficiently and flexibly because they ignore some practical factors, such as topic relationship, operating limits, and so on. In view of this, we devoted ourselves to improving real-time and reliable performance of the topic-based distributed publish/subscribe system based on WSN.This paper aims to take effective measures upon forwarding nodes in systems to improve the speed of forwarding messages while keeping the reliability. The message forwarding schema includes two parts:one is reading the message and matching to find routing information, which focuses on the system’s real-time performance; the other is forwarding message correctly which focuses on the system’s reliability.To improve the real-time performance, we need to get the routing information quickly from routing table. In this process, we need frequent operations, such as matching and query. So we use tree-structure topic clusters to arrange amount of structured topics and reduce the match time for routing and query. Meanwhile, the usage of topic clusters saves memory and makes the system more flexible. One of the contributions of this paper is establishing a policy repository to store policy items so that publishers can limit subscribers’operations indirectly. The policy repository can not only reduce unnecessary information transmission and relieve system pressure, but also group system nodes based on both location and logic level. The policy repository provides right routing information for lookup operation.We uses asynchronous parallel communication framework in order that forwarding nodes can transmit messages effectively and timely. In order to ensure the reliability of message forwarding, we use feedback mechanism to adjust transmission speed and make real-time data be delivered proportionately according to their priorities. Finally, we also do experiments to test the performance of the schema. The results show that it works well.